Overview of data collected in the field season of 2024 for coastal monitoring

Summary of 2024

Year in numbers

Number of detectors: 50

Number of detectors that survived the season: 48

Reasons of detector failure: Water damage by humans, seawater damage by other forces

Date of first deployment: 2024-06-14 11:46:00

Date of last deployment: 2024-07-20 08:33:00

Date of first retrieval: 2024-08-16 12:11:03

Date of last retrieval: 2024-10-17 16:30:00

Days detectors were on the field: 4908

Days with missing data: 318

% of missing data from all data: ca 6,5

Reasons of missing data:

1) No bats, no noise detected (later Nodata): 19

2) SD card full: 238

3) Batteries empty: 23

4) Detector dead: 33

Average time detector deployed: 91.84

Deployment times

This is a graph that shows the days with missing data and how long the detectors were in the field.

Noise

Figure 2 Shows the proportion of noise in all the sites. We can see that some of the sites are a lot noisier than others. Coastal sites tend to have more noise, but it varies from 20% (CM-27, sheltered by Kvassheim lighthouse) to 99% (Utsira-2). From lakesites, the most noise was 78% (CM-14, lake in Lista) and the least 11% (CM-34, guitarmuseum).

Comparison of noise in different sites

Figure 3

If we look at the total amount of 5 second recordings on Figure 4, then the amount of noise varies less, with CM-12 (coast near Lista) being an extra noisy location. The best overview of how locations differ gives Figure 5. We can see that the number of recordings we can work with is times bigger in lake sites than in coastal sites.

In terms of how to distribute sites to different classes, I have made a third, “inland” class that doesn’t really fit under the two others, with CM-01 (Utsira lighthouse), CM-07 (Raulemyra) and CM-41 (NMBU campus). Coastal sites are sites up to 400m from the coast, the furthest away being CM-25 near Brusand with 380 meters. There is also one lake site that falls into the 400m cutoff, CM-34 (Guitarmuseum) that is 280m from the coast. As for the inland sites, CM-01 is 600m from the coast and 480m from a freshwater lake, CM-07 is 9000m from the coast and 230m from a bigger freshwater source (wider part of a stream), CM-41 is 2230m from the coast and 100m from a bigger freshwater source (wider part of a stream).

Figure 4: Amount of recordings and noise

Figure 5: Correlation between recordings and the amount of noise

AUTO-ID bats

While the combined auto-id Figure 6 doesn’t give us a very detailed picture of what species are present, we can still get a proxy of the amount of recordings and species from it. When we compare the coast and the lakes, we can see that the amount of calls differs a lot (Figure 7).
Out of the top 4 coastal locations, top 2 are next to big rivers (Obrestad Prestegård, Brusand), third one is a fjord that looks like a lake (Framvaren) and fourth one is very sheltered from the sea (Stapnes bedehus).
CM-46 is the only lake site in a wind farm, but the low number of detections may be due to other reasons.

Timing of calls

We can see on Figure 8 that in September, the total amount of calls recorded starts to drop. We have to bear in mind here that the beginning and end of the graph is lower because the detectors were put up and taken down not all at once.

Northern bat activity

We can see a high level of avtivity throughout the recording period up until the beginning of september, when activity suddenly drops.

Nathusius’ pipistrelle activity

We can see that in our sites, the activity peaks in September.

Soprano pipistrelle activity

Activity is higher leading up to the September peak, but we can still clearly see a peak that occurs earlier than that of Nathusius’ pipistrelles.

Daubentons’ bat activity

Nathusius’ pipistrelle distribution

On the study area, some sites had clearly higher amounts of nathusius’ pipistrelles. On the Figure 16 below, we can see how many 5s clips of Auto-ID Pnat we had per site.

Figure 16: Number of Pnat calls per site

The median date of activity did not differ that much based on the location, as is seen on Figure 17 below.

Figure 17: Median date of Pnat calls per site